VOIDBENCHED: a hobbyist's silly LLM benchmarks, starting with FlirtBench
VoidStateKate · x · 2026-10-28
VoidStateKate launched VOIDBENCHED, a collection of non-serious benchmarks built from weird prompting styles and hilarious model outputs. After two years of exploring model personalities via unique prompts, her first bench is FlirtBench — testing which models can actually flirt — alongside 'model interviews' and more.
More from Fun
- Engineer compares working with AI agents to being married — yawnxyz · 2026-10-28
- Stunning AI-generated concept art casts goddess Sekhmet walking a post-apocalyptic wasteland — DionysianAgent · 2026-10-28
- Grok Turns One Question Into 60 Tasks, User Vows to Ignore Them All — verrsane · 2026-10-28
- Developer builds a multiplayer pixel-art drawing game that runs inside Twitter — jaivin · 2026-10-28
- A partial neuron atlas for Talkie-1930 yields eerily literary self-reflection — lfschiavo · 2026-10-28
- AI video brings classic monsters to life with vintage horror flair — ogcanuckamerican · 2026-10-28